- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Dear Embedded Community experts;
We need assistance for an intermittent behavior of our embedded system consisting of Intel Xeon-D 1559 based SBC’s running VxWorks-7 SR640. System handles average data size of 1 GBytes/sec from FPGA and thru PCIe switches.
SBC’s transfer data to each other using DMA. 1G Ethernet and RS-232 ports are also used for rare events.
VxWorks System Viewer records show that at intermittent intervals, all 12 cores of the CPU remain inactive, for a duration of approximately 3 milliseconds or 28 milliseconds. 28 msec events are rare than 3 msec. Number of occurences increase with increasing PCIe traffic. During this period Tick Timer interrupts are not served by OS either. It seems that DMA Engine or something else is suspending execution and a well-defined timeout mechanism is ending this locked state. 3 msec and 28 msec events are random in the order of 10-15 minutes or sometimes 1-8 hours.
We could not identified the mechanism that causes such a behavior.
What may cause dma engine to hang during host memory read and write to pcie transmit buffers? Do 3 or 28 ms durations adress any timeout setting?
Any insights, suggestions, or expertise provided would be greatly appreciated. Please feel free to request additional information or details if needed to aid in diagnosing and resolving this issue.
Thank you for your attention and support.
Warm regards,
Gursel Sahin
We need assistance for an intermittent behavior of our embedded system consisting of Intel Xeon-D 1559 based SBC’s running VxWorks-7 SR640. System handles average data size of 1 GBytes/sec from FPGA and thru PCIe switches.
SBC’s transfer data to each other using DMA. 1G Ethernet and RS-232 ports are also used for rare events.
VxWorks System Viewer records show that at intermittent intervals, all 12 cores of the CPU remain inactive, for a duration of approximately 3 milliseconds or 28 milliseconds. 28 msec events are rare than 3 msec. Number of occurences increase with increasing PCIe traffic. During this period Tick Timer interrupts are not served by OS either. It seems that DMA Engine or something else is suspending execution and a well-defined timeout mechanism is ending this locked state. 3 msec and 28 msec events are random in the order of 10-15 minutes or sometimes 1-8 hours.
We could not identified the mechanism that causes such a behavior.
What may cause dma engine to hang during host memory read and write to pcie transmit buffers? Do 3 or 28 ms durations adress any timeout setting?
Any insights, suggestions, or expertise provided would be greatly appreciated. Please feel free to request additional information or details if needed to aid in diagnosing and resolving this issue.
Thank you for your attention and support.
Warm regards,
Gursel Sahin
- Tags:
- DMA Completion timeout
- DMA Timeout
- Locked DMA
- Locked PCIe DMA
- Suspended DMA
- Suspended PCIe DMA
- Xeon D
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello @gSahin,
Thank you for contacting Intel Embedded Community.
Broadwell is not being supported anymore.
As you are using VxWorks, you may try asking to WindRiver support.
https://www.windriver.com/support
Best regards,
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page